No abstract

Wire bonding with bare Cu and Pd coated Cu (PdCu) wire have been adopted quickly as a mainstream packaging technology for high pin count and fine pitch devices. The differences between Au and Cu wire bonding are well understood as a result of extensive research. However, the differences between Cu and PdCu wire have not been investigated in as much detail. This paper is a result of collaborative work to study the wire bonding process using Cu wire and PdCu wire. 0.7 mil Cu and PdCu wires are used in the study with bonded ball diameter of about 33 µm. This is the leading edge of fine pitch Cu wire bonding. This study showed that Cu and PdCu wire can be bonded with similar 1st bond parameters, and the bonding results for these two wire types are similar. Process window test results with respect to the ultrasonic level and force level showed that shear, ball size and pad splash are very similar between Cu wire and PdCu wire. 1st bond pull strength is about 11% higher with PdCu wire indicating higher PdCu wire tensile strength. The after bake pull test results are very different between the two wires. PdCu wire has much higher pad peeling failure rate than Cu wire and the peeling showed up after 24 hour bake at 175ºC. The after bake pull strength is also much lower. The poor after bake results indicate that PdCu wire may need different bonding parameter settings than Cu wire. The effect of forming gas versus nitrogen should be examined as well. Second bond study showed the advantage of PdCu wire. The most significant difference is the tail pull strength. PdCu wire has 50% higher tail strength than Cu wire, indicating a more robust 2 nd bond process and less chance for short tails. A deeper understanding of these and other differences will assist in the proper selection between these wire types.High resolution transmission electron microscopy (HRTEM) with Energy Dispersive X-ray Spectrometer (EDX) studies show that the approximately 80 nm Pd coating dissolves into the Cu bulk during ball formation process, therefore, almost no Pd is present at the bond interface in the as-bonded state. However, Pd congregates and diffuses back to the bond interface, especially, a Pd-rich layer forms in the peripheral interface. The congregation of Pd near the bond interface appears to be detrimental to the bond strength, causing high peeling failure.

The ultrasonic friction power during thermosonic ball bonding with Au and Cu wires, both 25 µm in diameter, is derived with an improved method from experimental measurements during the bonding process. Experimental data include the current delivered to the ultrasonic transducer and the tangential force measured using piezoresistive microsensors integrated close to the Al bonding pad. The improvement results from a new, more accurate method to derive the mechanical compliance of the ultrasonic system. The method employs a bond process modification in which the ultrasonic current is ramped up sequentially in three steps. In the first two steps, the ultrasonic current is set to levels that are too low to cause sliding. The bonding takes place during the third step, when the current is ramped up to the optimum value required for making good quality bonds. The ultrasonic compliance values are derived from the first two steps and are 8.2 ± 0.5 µm N −1 and 7.7 ± 0.5 µm N −1 for the Au and Cu processes, respectively. These values are determined within an average error estimate of ±6%, substantially lower than the ±10% estimated with a previously reported method. The ultrasonic compliance in the case of Au is 6% higher due to the lower elastic modulus of Au compared with that of Cu. Typical maximum values of relative sliding amplitude of ultrasonic friction at the interface are 655 nm and 766 nm for the Au and Cu processes. These values are 81% of the free-air vibration amplitude of the bonding capillary tip for the respective ultrasonic current settings. Due to bond growth, which damps relative motion between the ball and the pad, the final relative amplitude at the bond interface is reduced to 4% of the equivalent free-air amplitude. Even though the maximum value of relative amplitude is 17% higher in the Cu process compared with the Au process, the average total interfacial sliding is 519 µm in the Cu process, which is 31% lower than that in the Au process (759 µm). The average maximum interfacial friction power is 10.3 mW and 16.9 mW for the Au and Cu ball bonding processes, respectively. The total sliding friction energy delivered to the bond is 48.5 µJ and 49.4 µJ for the Au and Cu ball bonding cases, respectively. These values result in average friction energy densities of 50.3 mJ mm −2 and 54.8 mJ mm −2 for Au and Cu ball bonding, respectively.

No abstract

No abstract